A revised time patch (fwd)
H.J. Lu
hjl@lucon.org
Tue Nov 3 20:54:00 GMT 1998
>
> hjl@lucon.org (H.J. Lu) writes:
>
> > Here is the testcase. My previous patch to time/mktime.c is incomplete.
> > Please use this one. BTW, __mktime_internal mishandles the cases
> > where tm_isdst is different from __daylight != 0. Basically when
> > tm_isdst == 0/__daylight != 0 and tm_isdst == 1/__daylight == 0, we
> > should check
>
> I've checked in a patch for this now. The test case HJ sent (thanks!)
> work now and I hopefully have't broken anything else.
>
Unfortunately, it is still broken. I am enclosing the expected output
from my testcase.
> I haven't used HJ's patch since the mktime file must work standalone
> in other packages and therefore must not depend on any implementation
> details of the other timezone functions.
>
What do you mean by that? Are we talking glibc here? If you really
want, you can #ifdef/#endif with __GLIBC__ >= 2. At least, we get
a working glibc. My patch is the cleanest one I can find for glibc.
I don't really care other packages.
Thanks.
H.J.
------
min: 59
hour: 1
isdst: 1
593917199
Thu Oct 27 01:59:59 1988
min: 0
hour: 1
isdst: 0
593917221
Thu Oct 27 01:00:21 1988
min: 0
hour: 3
isdst: 1
574999221
Tue Mar 22 03:00:21 1988
min: 59
hour: 1
isdst: 1
593913599
Thu Oct 27 01:59:59 1988
min: 0
hour: 4
isdst: 1
574999221
Tue Mar 22 04:00:21 1988
min: 0
hour: 0
isdst: 0
593913621
Thu Oct 27 00:00:21 1988
More information about the Libc-hacker
mailing list